Re: Corruption during WAL replay

On Tue, Aug 18, 2020 at 3:22 AM Andres Freund <andres(at)anarazel(dot)de> wrote:
>
> Hi,
>
> On 2020-08-17 14:05:37 +0300, Heikki Linnakangas wrote:
> > On 14/04/2020 22:04, Teja Mupparti wrote:
> > > Thanks Kyotaro and Masahiko for the feedback. I think there is a
> > > consensus on the critical-section around truncate,
> >
> > +1
>
> I'm inclined to think that we should do that independent of the far more
> complicated fix for other related issues.

+1

If we had a critical section in RelationTruncate(), crash recovery
would continue failing until the situation of the underlying file is
recovered if a PANIC happens. The current comment in
RelationTruncate() says it’s worse than the disease. But considering
physical replication, as Andres mentioned, a failure to truncate the
file after logging WAL is no longer a harmless failure. Also, the
critical section would be necessary even if we reversed the order of
truncation and dropping buffers and resolved the issue. So I agree to
proceed with the patch that adds a critical section independent of
fixing other related things discussed in this thread. If Teja seems
not to work on this I’ll write the patch.

Regards,

--
Masahiko Sawada
EnterpriseDB: https://www.enterprisedb.com/